TWI551119B - Video encoding apparatus and video decoding apparatus and encoding method and decoding method thereof - Google Patents

Description

Video coding device and video decoding device, and codec and decoding method thereof

The present invention is related to multimedia signal processing techniques and, in particular, to encoding/decoding techniques in video systems.

With the advancement of communication technology, digital TV broadcasting has become mature and popular. In addition to transmission via cable lines, digital TV signals can also be transmitted in the form of wireless signals via devices such as base stations or satellites. In order to balance the need to improve the picture quality and reduce the amount of data transferred, the transmitting end usually encodes and compresses the image and sound signals to be transmitted. Correspondingly, the receiving end must correctly decode and decompress the received signal to restore the video signal.

Figure 1 presents a partial functional block diagram of an encoding system that conforms to one of the audio and video coding standard (AVS). The intra-prediction module 12 is responsible for performing an intra-frame prediction process for each image block in a video frame to generate a luminance residual block of each image block. The luminance residual value block is then provided to a discrete cosine transform (DCT) module 14 for discrete cosine transform to produce a discrete cosine transform coefficient matrix. In order to further reduce the amount of data, the secondary transform module 16 applies a quadratic conversion to the low frequency components in the discrete cosine transform coefficient matrix. In the case of the AVS encoding system, regardless of the size of the discrete cosine transform coefficient matrix, the secondary conversion module 16 performs secondary conversion for the 4*4 low frequency components located at the uppermost left corner. Subsequently, after the second conversion of the low frequency components, and other non-secondary conversion The shifted high frequency discrete cosine transform coefficients are recombined at the quantization module 18 and a quantization procedure is applied.

In practice, the discrete cosine transform performed by the discrete cosine transform module 14 includes a set of discrete cosine transforms in the vertical direction and a set of discrete cosine transforms in the horizontal direction. Similarly, the secondary conversion performed by the secondary conversion module 16 is composed of a set of secondary conversions performed in the vertical direction and a set of secondary conversions performed in the horizontal direction. According to the AVS specification, the discrete cosine transform module 14 should first perform discrete cosine transform on the luminance residual value block in the horizontal direction until all discrete cosine transforms in the horizontal direction are completed, and then start to be discretely column by column in the vertical direction. Cosine conversion. On the other hand, according to the AVS specification, the secondary conversion module 16 should first perform secondary conversion on the low-frequency components in the discrete cosine transform coefficient matrix in the vertical direction by column until all secondary conversions in the vertical direction are completed, and then Start secondary conversion column by column in the horizontal direction. Figure 2 (A) presents the sequential relationship of each of the above transitions.

Taking the case where the size of the luminance residual value block is 4*4 as an example, FIG. 2(B) shows the detailed timing relationship of the typical discrete cosine transform and the secondary conversion in the AVS encoding system. The symbols R ₀ to R ₃ represent four columns in the block, and the symbols C ₀ to C ₃ represent four columns in the block. As shown in Fig. 2(B), in the working period 0~3, the discrete cosine transform in the horizontal direction is sequentially applied to the columns R ₀ to R ₃ ; in the working period 4~7, in the vertical direction The discrete cosine transform is applied sequentially to column C ₀ ~column C ₃ . After the discrete cosine transform applied to the column C ₀ is completed, the content of the column C ₀ is no longer changed by the influence of the discrete cosine transform, and the secondary conversion module 16 can thus face the column C ₀ in the vertical direction from the duty cycle 5. Apply a secondary conversion. And so on, in the working cycle 6~8, the secondary conversion in the vertical direction is sequentially applied to the column C ₁ ~ the column C ₃ ; in the working cycle 9~12, the secondary conversion in the horizontal direction is sequential Applied to column R ₀ ~ column R ₃ . As can be seen from Fig. 2(B), it takes a total of 13 duty cycles to complete the discrete cosine transform and the quadratic conversion for a 4*4 luminance residual block. In contrast, if a secondary conversion is not applied, it takes only 8 duty cycles to complete the discrete cosine transform for a 4*4 luminance residual block. That is to say, in a typical AVS encoding system, although the secondary conversion can contribute to the benefit of reducing the amount of data of the encoding result, it will cause an extension of the encoding operation time, thereby affecting the overall efficiency of the system.

In order to solve the above problems, the present invention provides a new video encoding device and video decoding device. By combining discrete cosine transform and quadratic conversion with appropriate job scheduling, the number of duty cycles required for video encoding/decoding can be effectively reduced.

According to an embodiment of the invention, a video encoding device includes a conversion module. The conversion module is configured to apply a combined conversion to a specific image data block along a specific direction according to a conversion matrix. The conversion matrix is the product of a preliminary conversion matrix and a secondary transformation matrix. The preliminary conversion matrix corresponds to a one-dimensional preliminary conversion in the particular direction in a two-dimensional preliminary transformation. The quadratic conversion matrix corresponds to one-dimensional quadratic conversion in the specific direction in a two-dimensional quadratic conversion.

Another embodiment of the present invention is a video encoding method. The video encoding method includes a converting step of applying a combined conversion to a specific image data block in a specific direction according to a conversion matrix. The conversion matrix is the product of a preliminary conversion matrix and a secondary transformation matrix. The preliminary conversion matrix corresponds to a one-dimensional preliminary conversion in the particular direction in a two-dimensional preliminary transformation. The quadratic conversion matrix corresponds to one-dimensional quadratic conversion in the specific direction in a two-dimensional quadratic conversion.

Another embodiment of the present invention is a video decoding device including a reverse conversion module. The inverse conversion module is configured to apply a combined reverse conversion to a specific image data block according to a reverse conversion matrix. The inverse transformation matrix is the product of a reverse preliminary transformation matrix and an inverse quadratic transformation matrix. The reverse preliminary conversion matrix corresponds to a one-dimensional inverse preliminary conversion in the specific direction in a two-dimensional reverse preliminary conversion. The inverse quadratic conversion matrix corresponds to one-dimensional inverse quadratic conversion in the specific direction in a two-dimensional inverse quadratic conversion.

Another embodiment of the present invention is a video decoding method. The video The decoding method includes a reverse conversion step of applying a combined back-conversion to a specific image data block in a specific direction according to a reverse conversion matrix. The inverse transformation matrix is the product of a reverse preliminary transformation matrix and an inverse quadratic transformation matrix. The reverse preliminary conversion matrix corresponds to a one-dimensional inverse preliminary conversion in the specific direction in a two-dimensional reverse preliminary conversion. The inverse quadratic conversion matrix corresponds to one-dimensional inverse quadratic conversion in the specific direction in a two-dimensional inverse quadratic conversion.

The advantages and spirit of the present invention will be further understood from the following detailed description of the invention.

12‧‧‧ In-frame prediction module

14‧‧‧Discrete cosine transform module

16‧‧‧Secondary conversion module

18‧‧‧Quantitative Module

300‧‧‧Video coding device

32‧‧‧Intraframe prediction module

34‧‧‧Transition module

38‧‧‧Quantitative Module

S51~S52‧‧‧ Process steps

S61~S64‧‧‧ Process steps

S71~S74‧‧‧ Process steps

Figure 1 presents a partial functional block diagram of an encoding system that conforms to one of the Digital Audio and Video Codec Technical Standards (AVS).

Figure 2 (A) and Figure 2 (B) present the timing relationship of discrete cosine transform / quadratic conversion in accordance with one of the digital audio and video codec standards (AVS).

Figure 3 is a functional block diagram of a video encoding apparatus in accordance with an embodiment of the present invention.

Figure 4 (A) ~ Figure 4 (D) show the four relative relationship possibilities of the secondary conversion size and the initial conversion size.

Figures 5(A) to 5(C) show examples of operational procedures of a conversion module in accordance with an embodiment of the present invention.

6(A) and 6(B) show examples of operational procedures of a conversion module in accordance with another embodiment of the present invention.

7(A) and 7(B) show examples of operational procedures of a conversion module in accordance with yet another embodiment of the present invention.

8(A) and 8(B) show examples of operational procedures of a conversion module in accordance with an embodiment of the present invention.

9(A) and 9(B) show examples of operational procedures of a conversion module in accordance with another embodiment of the present invention.

It should be noted that the drawings of the present invention include functional block diagrams that present a plurality of functional modules associated with each other. These figures are not detailed circuit diagrams, and the connecting lines therein are only used to represent the signal flow. Multiple interactions between functional components and/or procedures do not have to be achieved through direct electrical connections. In addition, the functions of the individual components are not necessarily allotted in the manner illustrated in the drawings, and the decentralized blocks are not necessarily implemented in the form of decentralized electronic components.

A video encoding apparatus according to an embodiment of the present invention. In practice, the video encoding device can be integrated in various video encoding systems with a secondary transform mechanism, or can exist independently. For convenience of description, the following description mainly takes the case where the video encoding device is installed in a digital audio and video codec technical standard (AVS) encoding system, and a partial functional block diagram thereof is shown in FIG.

In this embodiment, the intra-frame prediction module 32 is configured to perform an intra-frame prediction process for each image block in a video frame to generate a luminance residual value block of each image block. The conversion module 34 is responsible for applying a preliminary conversion and a second conversion to the luminance residual value block. For example, the preliminary conversion may be a discrete cosine transform (DCT) or an integer transform transformed by a discrete cosine transform, but is not limited thereto. Since the secondary conversion is a low frequency component applied to the preliminary conversion result, the size of the secondary conversion (the range size of the applied object) is necessarily smaller than or equal to the size of the preliminary conversion. Figure 4 (A) ~ Figure 4 (D) illustrate the four relative relationship possibilities of the secondary conversion size and the initial conversion size. The outer frame limit represents the size of the preliminary conversion result, and the area filled with the black oblique line represents the application range of the secondary conversion. In these legends, the preliminary conversion result contains N ₁ column, N ₂ column elements, and the secondary conversion will be applied to the M ₁ column, M ₂ column elements.

First, it is explained how the conversion module 34 operates in the case where the two-dimensional preliminary conversion and the two-dimensional secondary conversion are the same as shown in FIG. 4(D) (M ₁ =N ₁ and M ₂ =N ₂ ). Since both the two-dimensional preliminary conversion and the two-dimensional secondary conversion are linear transformations, changing the order in which the conversion is performed does not affect the final conversion result. Therefore, as shown in FIG. 5(A), the horizontal column-by-column secondary conversion can be advanced to the horizontal column-by-column preliminary conversion and the vertical column-by-column preliminary conversion. On the other hand, if M ₁ =N ₁ and M ₂ =N ₂ , the horizontal column-by-column preliminary conversion and the horizontal column-by-column secondary conversion can be combined into a single conversion (step S51), and the vertical column-by-column preliminary conversion can also be The vertical column-by-column secondary conversion is combined into a single conversion (step S52). More specifically, the conversion matrix corresponding to the horizontal column-by-column preliminary conversion is multiplied by the conversion matrix corresponding to the horizontal column-by-column secondary conversion, that is, the conversion matrix converted horizontally and column-by-column. Similarly, the conversion matrix corresponding to the vertical column-by-column preliminary conversion is multiplied by the conversion matrix corresponding to the vertical column-by-column secondary conversion, that is, the conversion matrix which is converted vertically and column-by-column. Figure 5 (B) takes the case of M ₁ = N ₁ = M ₂ = N ₂ = 4 as an example, and presents a detailed operational timing example of the conversion module 34 in this case. In the working cycle 0~3, the post-bonding conversion in the horizontal direction is sequentially applied to the columns R ₀ to R ₃ ; in the working cycles 4 to 7, the post-combination conversion in the vertical direction is sequentially applied to the column C. ₀ ~ column C ₃ . As can be seen from FIG. 5(B), the conversion module 34 only needs 8 working cycles to complete the two-dimensional preliminary conversion and the two-dimensional secondary conversion, which is obviously more efficient than the conventional method. Subsequently, the output signal of the conversion module 34 is supplied to the quantization module 36 for quantization.

In the Digital Audio Video Codec Standard (AVS) coding system, when M ₁ = N ₁ = M ₂ = N ₂ = 4, the combined conversion matrix is:

In practice, the conversion module 34 can be implemented to include fixed and/or programmable digital logic circuits, such as programmable logic gate arrays, application specific integrated circuits, microcontrollers, microprocessors, digital signal processors. , with other necessary circuits. It should be noted that according to a known The technique of applying a conversion to a video data block by a specific conversion matrix is known to those of ordinary skill in the art to which the present invention pertains, and will not be described herein.

FIG. 5(C) presents another operational timing example of the conversion module 34. In this example, the conversion module 34 only combines the horizontal column-by-column preliminary conversion and the horizontal column-by-column secondary conversion, and after the combined post-transformation in the horizontal direction is completed, the vertical column-by-column preliminary conversion and the vertical column-by-column are respectively performed. Conversion. If such an operation sequence is adopted, the conversion module 34 only needs 9 working cycles to complete the two-dimensional preliminary conversion and the two-dimensional secondary conversion, which is more efficient than the conventional method.

In some cases, when a target image data block is determined not to perform one-dimensional secondary conversion in the horizontal direction, the above-described conversion matrix corresponding to the horizontal column-by-column secondary conversion may be set as an identity matrix. The transformation matrix that makes the horizontal column-by-column transformation is directly equal to the transformation matrix of the horizontal column-by-column preliminary transformation. Similarly, when a certain target image data block is determined not to perform one-dimensional secondary conversion in the vertical direction, the above-mentioned conversion matrix corresponding to the vertical column-by-column secondary conversion may be set as a unit matrix, so that the vertical The conversion matrix after the column combination is directly equal to the conversion matrix of the vertical column-by-column conversion.

Next, how the conversion module 34 operates in the case where the two-dimensional preliminary conversion and the two-dimensional secondary conversion as shown in FIG. 4(C) are different in size (M ₁ =N ₁ and M ₂ <N ₂ ). FIG six (A), in step S61, the conversion module 34 first be subjected to one-dimensional in the vertical direction for all the data column by column (Column 1 Column M - ₁₎ in the input image data block Initial conversion. The resulting preliminary conversion result can be considered to contain two sub-blocks: the first sub-block is the first M ₂ column in the preliminary conversion result, and the second sub-block is after the preliminary conversion result ( N ₂ -M ₂ ) column. The first sub-block is the area that needs to be subjected to the secondary conversion, and the second sub-block is no. The conversion module 34 may perform horizontal column-by-column combining conversion after the first sub-block according to the combined conversion matrix (step S62), and may perform horizontal-by-column one-dimensional preliminary conversion on the second sub-block at the same time (step S63). ). At this point, the horizontal and vertical preliminary conversion of the second sub-block has been completed. Then, in step S64, the conversion module 34 only needs to perform the vertical column-by-column one-dimensional secondary conversion on the intermediate conversion result generated in step S62. Figure 6 (B) takes the case of M ₁ = N ₁ = N ₂ = 4 and M ₂ = 2 as an example, and presents a detailed operational timing example of the conversion module 34 in this case. In the work cycle 0~3, the one-dimensional preliminary conversion in the vertical direction is sequentially applied to the column C ₀ to the column C ₃ . In the working cycle 4~5, the horizontal column-by-column conversion system is sequentially applied to the column R ₀ and the column R ₁ , and the horizontal column-by-column preliminary conversion system is sequentially applied to the column R ₂ and the column R ₃ . In the work cycle 6~9, the vertical column-by-column secondary conversion is sequentially applied to the first two columns of columns C ₀ to C ₃ .

In the case where the two-dimensional preliminary conversion and the two-dimensional secondary conversion are different as shown in FIG. 4(B) (M ₁ <N ₁ and M ₂ =N ₂ ), an example of the operation mode of the conversion module 34 is shown. Figure 7 (A). In step S71, the conversion module 34 first for all block data in the input image data column by column (column 1 of row M ₂₎ applied in the horizontal direction one-dimensional initial conversion. The resulting preliminary conversion result can be considered to contain two sub-blocks: the first sub-block is the first M ₁ column in the preliminary conversion result, and the second sub-block is after the preliminary conversion result ( N ₁ -M ₁ ) column. The first sub-block is the area that needs to be subjected to the secondary conversion, and the second sub-block is no. The conversion module 34 can perform vertical column-by-column combination conversion on the first sub-block according to the combined conversion matrix (step S72), and can simultaneously perform vertical column-by-column one-dimensional preliminary conversion on the second sub-block (step S73). ). At this point, the horizontal and vertical preliminary conversion of the second sub-block has been completed. Then, in step S74, the conversion module 34 only needs to perform horizontal column-by-column one-dimensional secondary conversion on the intermediate conversion result generated in step S72. Figure 6 (B) takes the case of M ₂ = N ₁ = N ₂ = 4 and M ₁ = 2 as an example, and presents a detailed operational timing example of the conversion module 34 in this case. In the duty cycle 0~3, the one-dimensional preliminary conversion in the horizontal direction is sequentially applied to the columns R ₀ to R ₃ . In the work cycle 4~5, the vertical column-by-column combination conversion system is sequentially applied to the column C ₀ and the column C ₁ , and the vertical column-by-column preliminary conversion system is sequentially applied to the column C ₂ and the column C ₃ . In the work cycle 6~9, the horizontal column-by-column secondary conversion is sequentially applied to the first two columns of the columns R ₀ to R ₃ .

When the dimensions of the two-dimensional quadratic transformation in both the horizontal and vertical directions are smaller than the two-dimensional preliminary transformation (that is, the case of M ₁ <N ₁ , M ₂ <N ₂ as shown in FIG. 4(A)), The number of work cycles required for the initial conversion, the number of work cycles required for the second conversion is small, and the overall efficiency is not greatly affected. The conversion module 34 can operate according to the process illustrated in Figure 2 (A), and does not necessarily have to be combined. After conversion.

If the video coding system cooperated with the conversion module 34, the two-dimensional preliminary conversion And the size of the two-dimensional secondary conversion is fixed, and the conversion module 34 can be designed to be fixed by one of the operating procedures shown in FIG. 5(A), FIG. 6(A) or FIG. 7(A). . In contrast, in a video coding system in which the size of the two-dimensional preliminary conversion and the two-dimensional secondary conversion may dynamically change, the conversion module 34 can be designed to select its operational procedure as appropriate.

Another embodiment of the present invention is a video encoding method. The video encoding method includes a converting step of applying a combined conversion to a specific image data block in a specific direction according to a conversion matrix. The conversion matrix is the product of a preliminary conversion matrix and a secondary transformation matrix.

The preliminary conversion matrix corresponds to a one-dimensional preliminary conversion in the particular direction in a two-dimensional preliminary transformation. The quadratic conversion matrix corresponds to one-dimensional quadratic conversion in the specific direction in a two-dimensional quadratic conversion. The various operational changes previously described in the introduction of the video encoding device 300 (e.g., further performing post-combination conversion perpendicular to the particular direction) may also be applied to the video encoding method, the details of which are not described again.

The concept of the present invention is applicable not only to various video coding systems having a secondary conversion mechanism, but also to video decoding systems having a reverse secondary conversion mechanism. More specifically, the two-dimensional reverse preliminary conversion includes a set of reverse preliminary transitions in the vertical direction and a set of reverse preliminary transitions in the horizontal direction. Similarly, two-dimensional inverse quadratic transformation consists of a set of inverse quadratic transformations performed in the vertical direction and a set of inverse quadratic transformations performed in the horizontal direction. When the two-dimensional reverse preliminary conversion and the two-dimensional inverse secondary conversion are both linear transformations, changing the order in which the conversion is performed does not affect the final conversion result. Therefore, the one-dimensional reverse preliminary conversion and the one-dimensional inverse secondary conversion in the same direction can be selectively combined to improve the decoding efficiency.

According to an embodiment of the invention, a video decoding apparatus includes a reverse conversion module. The inverse conversion module is configured to apply a combined reverse conversion to a specific image data block according to a reverse conversion matrix. The inverse transformation matrix is the product of a reverse preliminary transformation matrix and an inverse quadratic transformation matrix. Reverse initial conversion matrix pair A one-dimensional reverse preliminary conversion should be performed in this particular direction in a two-dimensional reverse preliminary conversion. The inverse quadratic conversion matrix corresponds to one-dimensional inverse quadratic conversion in the specific direction in a two-dimensional inverse quadratic conversion. The various operational changes previously described in the introduction of the video encoding device 300 may also be correspondingly applied to the video decoding device (e.g., further performing a combination of back-conversions perpendicular to the particular direction), the details of which are not described again. It should be noted that even if the encoding end does not employ the post-combination conversion according to the present invention, the decoding end may employ the combined back-conversion according to the present invention.

In the AVS decoding system, when the two-dimensional reverse preliminary conversion and the two-dimensional inverse secondary conversion are both 4*4, the combined inverse conversion matrix is:

Another embodiment of the present invention is a video decoding method. The video decoding method includes a reverse conversion step of applying a combined reverse conversion to a specific image data block in a specific direction according to a reverse conversion matrix. The inverse transformation matrix is the product of a reverse preliminary transformation matrix and an inverse quadratic transformation matrix. The reverse preliminary conversion matrix corresponds to a one-dimensional inverse preliminary conversion in the specific direction in a two-dimensional reverse preliminary conversion. The inverse quadratic conversion matrix corresponds to one-dimensional inverse quadratic conversion in the specific direction in a two-dimensional inverse quadratic conversion.

According to another embodiment of the present invention, a video encoding apparatus includes a conversion module that is responsible for applying a preliminary conversion and a second conversion to the image data block. In this embodiment, the conversion module appropriately schedules the one-dimensional preliminary conversion and the one-dimensional secondary conversion in the same direction to achieve an effect of improving coding efficiency. As shown in FIG. 8(A), the conversion module advances the horizontal column-by-column secondary conversion to the horizontal column-by-column preliminary conversion and the vertical column-by-column preliminary conversion. It should be noted that after completing the initial conversion of the first column level, the conversion module immediately starts a horizontal secondary conversion column by column. After completing all horizontal secondary conversions, the conversion module begins vertical vertical conversion by column. After completing the initial vertical conversion of the first column, the conversion module immediately starts a vertical secondary conversion by column. Taking the case where the two-dimensional preliminary conversion and the two-dimensional secondary conversion are all 4*4 as an example, FIG. 8(B) shows the detailed operation timing of the conversion module. In the duty cycle 0~3, the preliminary conversion in the horizontal direction is sequentially applied to the columns R ₀ to R ₃ . In the duty cycle 1 to 4, the secondary conversion in the horizontal direction is sequentially applied to the columns R ₀ to R ₃ . Subsequently, in the work cycles 5 to 8, the preliminary conversion in the vertical direction is sequentially applied to the columns C ₀ to C ₃ . In the work cycles 6 to 9, the secondary conversion in the vertical direction is sequentially applied to the columns C ₀ to C ₃ . It can be seen from Fig. 8(B) that it takes only 10 working cycles to complete the initial conversion and the secondary conversion of a 4*4 image data block.

Similarly, the case where the size of the two-dimensional preliminary conversion and the two-dimensional secondary conversion are both 4*4 is taken as an example, and FIG. 9(A) and FIG. 9(B) present another detailed operational timing of the conversion module according to the present invention. In this example, the conversion module advances the vertical column-by-column preliminary conversion and the vertical column-by-column secondary conversion to the horizontal column-by-column preliminary conversion and the horizontal column-by-column secondary conversion. As shown in FIG. 9(B), in the duty cycle 0 to 3, the preliminary conversion in the vertical direction is sequentially applied to the column C ₀ to the column C ₃ . In the duty cycle 1 to 4, the secondary conversion in the vertical direction is sequentially applied to the column C ₀ to the column C ₃ . Subsequently, in the duty cycle 5-8, the preliminary conversion in the horizontal direction is sequentially applied to the columns R ₀ to R ₃ . In the duty cycle 6 to 9, the secondary conversion in the horizontal direction is sequentially applied to the columns R ₀ to R ₃ . It can be seen from Fig. 9(B) that it takes only 10 working cycles to complete the initial conversion and the secondary conversion of a 4*4 image data block.

According to another embodiment of the present invention, a video encoding method includes the following steps for: (a) performing a horizontal preliminary conversion column by column; (b) after completing the initial conversion of the first column level, Immediately start a horizontal secondary conversion column by column; (c) perform a vertical preliminary conversion by column; and (d) start a vertical secondary conversion by column after completing the vertical conversion of the first column. Step (c) is arranged to start after step (b) is completely completed, or step (a) is arranged to start after step (d) is completely completed.

The above-mentioned practice of preliminary conversion/secondary conversion scheduling also applies to a video decoding system having a reverse secondary conversion mechanism. Another embodiment of the present invention is a video solution The code device includes a reverse conversion module. The reverse conversion module is configured to perform a horizontal reverse initial conversion for an image data block column by column, and immediately start a horizontal reverse secondary conversion column by column after completing the first column horizontal reverse preliminary conversion. . The reverse conversion module also performs a vertical reverse preliminary conversion by column, and immediately after the initial vertical reverse conversion of the first column, a vertical reverse secondary conversion is started. It should be noted that the reverse conversion module starts the vertical reverse initial conversion after completing all horizontal reverse secondary conversions, or starts the horizontal reverse preliminary after completing all vertical reverse secondary conversions. Conversion.

Another embodiment of the present invention is a video decoding method, including the following steps for an image data block: (a) performing a horizontal reverse preliminary conversion column by column; (b) performing a horizontal inversion of the first column. Immediately after the initial conversion, a horizontal reverse quadratic conversion is started column by column; (c) a vertical reverse initial conversion is performed column by column; and (d) the column is immediately started after the initial vertical inversion of the first column is completed. Perform a vertical reverse quadratic conversion. Wherein step (c) is arranged to start after step (b) is completely completed, or step (a) is arranged to start after step (d) is completely completed.

The features and spirit of the present invention will be more apparent from the detailed description of the preferred embodiments. On the contrary, the intention is to cover various modifications and equivalents within the scope of the invention as claimed.

S51~S52‧‧‧ Process steps

Claims (32)

A video coding device includes: a conversion module, configured to perform a combined transformation on a target image data block along a specific direction according to a conversion matrix, wherein the conversion matrix is a preliminary conversion matrix and a second time Transforming a product of a matrix, and the preliminary transformation matrix corresponds to a one-dimensional preliminary transformation in the two-dimensional preliminary transformation along the specific direction, the secondary transformation matrix corresponding to one of the two-dimensional secondary transformations in the specific direction Dimensional secondary transformation. The video encoding device of claim 1, wherein the conversion module is further configured to: apply another combined conversion to the target image data block perpendicular to the specific direction according to another conversion matrix, Wherein the another conversion matrix is a product of another preliminary conversion matrix and another secondary transformation matrix, and the another preliminary conversion matrix corresponds to one-dimensional initial conversion in the two-dimensional preliminary transformation perpendicular to the specific direction, Another quadratic transformation matrix corresponds to one-dimensional quadratic conversion perpendicular to the particular direction in the two-dimensional quadratic transformation. The video encoding device of claim 1, wherein the two-dimensional preliminary conversion is the same as the size of the two-dimensional secondary conversion. The video encoding device of claim 1, wherein the conversion matrix is: . The video encoding device according to claim 1, wherein the video encoding device is applied to an encoding system of a digital audio and video codec technology standard (AVS). The video encoding device according to claim 1, wherein the target image data area The block is determined not to perform one-dimensional secondary conversion in the specific direction, and the secondary conversion matrix is set as an identity matrix. The video encoding device of claim 1, wherein the two-dimensional preliminary conversion is perpendicular to the specific direction and the size is greater than the size of the two-dimensional secondary conversion perpendicular to the specific direction; the conversion module is further used Before performing the post-combination conversion, an input image data block is subjected to one-dimensional initial conversion in the two-dimensional preliminary conversion in the vertical direction to generate a first sub-block and a second sub-area. a preliminary conversion result of the block, wherein the first sub-block is the target image data block, and the size of the first sub-block perpendicular to the specific direction is that the two-dimensional secondary conversion is perpendicular to the specific direction Dimensions; performing a one-dimensional initial conversion along the specific direction in the two-dimensional preliminary transformation on the second sub-block; and performing an intermediate conversion on the target image data block to be subjected to the combined conversion As a result, one-dimensional secondary conversion is performed perpendicular to the specific direction in the two-dimensional quadratic conversion. A video encoding method includes: applying a combined transformation to a specific image data block along a specific direction according to a conversion matrix, wherein the conversion matrix is a product of a preliminary conversion matrix and a secondary conversion matrix, and the The preliminary conversion matrix corresponds to a one-dimensional preliminary transformation along the specific direction in a two-dimensional preliminary transformation, and the secondary transformation matrix corresponds to one-dimensional secondary transformation in the specific direction in a two-dimensional secondary transformation. The video encoding method of claim 8, further comprising: applying another combined conversion to the target image data block perpendicular to the specific direction according to another conversion matrix, wherein the another conversion matrix is a product of another preliminary conversion matrix and another secondary transformation matrix, and the another preliminary conversion matrix corresponds to one-dimensional initial conversion perpendicular to the specific direction in the two-dimensional preliminary transformation, and the other second transformation The commutation matrix corresponds to one-dimensional quadratic conversion perpendicular to the particular direction in the two-dimensional quadratic transformation. The video encoding method of claim 8, wherein the two-dimensional preliminary conversion is the same as the size of the two-dimensional secondary conversion. The video encoding method according to claim 10 is applied to a digital audio and video codec technical standard (AVS), wherein the conversion matrix is: . The video encoding method of claim 8, wherein the secondary image conversion matrix is set to a unit matrix when the target image data block is determined not to perform one-dimensional secondary conversion in the specific direction. . The video encoding method of claim 8, wherein the two-dimensional preliminary conversion is perpendicular to the size of the specific direction and the size of the two-dimensional secondary conversion is perpendicular to the specific direction; the video encoding method further comprises: Before performing the post-combination conversion, an input image data block is subjected to a one-dimensional preliminary conversion in the two-dimensional preliminary conversion in the vertical direction to generate a first sub-block and a second sub-block. a preliminary conversion result, wherein the first sub-block is the target image data block, and the size of the first sub-block perpendicular to the specific direction is the size of the two-dimensional secondary conversion perpendicular to the specific direction Transmitting, by the second sub-block, a one-dimensional initial conversion in the specific direction in the two-dimensional preliminary transformation; and performing an intermediate conversion result on the target image data block to be subjected to the combination conversion, The two-dimensional quadratic conversion is applied to perform one-dimensional secondary conversion perpendicular to the specific direction. A video decoding device includes: a reverse conversion module, configured to perform a combined reverse conversion on a target image data block according to a reverse conversion matrix, wherein the reverse conversion matrix is a product of a reverse preliminary transformation matrix and an inverse quadratic transformation matrix, and the inverse preliminary transformation matrix corresponds to a one-dimensional inverse preliminary transformation in the specific direction in a two-dimensional inverse preliminary transformation, the reverse The quadratic transformation matrix corresponds to one-dimensional inverse quadratic transformation along the particular direction in a two-dimensional inverse quadratic transformation. The video decoding device of claim 14, wherein the inverse conversion module is further configured to: apply another vertical direction to the target image data block according to another reverse conversion matrix a combined inverse transform, wherein the other inverse transform matrix is a product of another inverse preliminary transform matrix and another inverse quadratic transform matrix, and the another reverse preliminary transform matrix corresponds to the two-dimensional inverse Performing a one-dimensional inverse preliminary transformation perpendicular to the specific direction in the preliminary conversion, the another inverse quadratic transformation matrix corresponding to the one-dimensional inverse of the two-dimensional inverse quadratic transformation perpendicular to the specific direction Sub-conversion. The video decoding device of claim 14, wherein the two-dimensional reverse preliminary conversion is the same size as the two-dimensional inverse secondary conversion. The video decoding device of claim 14, wherein the inverse conversion matrix is: . The video decoding device according to claim 14, wherein the video decoding device is applied to a decoding system of a digital audio and video codec technology standard (AVS). The video decoding device of claim 14, wherein the target image data block is determined not to perform one-dimensional inverse quadratic conversion in the specific direction, the reverse second The conversion matrix is set to a unit matrix. The video decoding device of claim 14, wherein the two-dimensional reverse preliminary conversion is perpendicular to the size of the specific direction greater than the size of the two-dimensional inverse quadratic transformation perpendicular to the specific direction; the reverse conversion The module is further configured to: perform a one-dimensional reverse initial conversion in the two-dimensional reverse preliminary conversion on the input image data block before performing the combined back-conversion to generate the inclusion a preliminary conversion result of one of the first sub-block and the second sub-block, wherein the first sub-block is the target image data block, and the first sub-block is perpendicular to the size of the specific direction The first two-dimensional inverse quadratic transformation is perpendicular to the size of the specific direction; the second sub-block is subjected to one-dimensional reverse preliminary conversion in the two-dimensional reverse preliminary conversion in the specific direction; The target image data block is subjected to one of the intermediate conversion results generated by the inverse transformation after the combination, and the one-dimensional inverse quadratic conversion is performed perpendicular to the specific direction in the two-dimensional inverse quadratic conversion. A video decoding method includes: applying a combined inverse transform to a target image data block according to a reverse transform matrix, wherein the inverse transform matrix is a reverse preliminary conversion matrix and a reverse a product of a quadratic transformation matrix, and the inverse preliminary transformation matrix corresponds to a one-dimensional inverse preliminary transformation in the two-dimensional inverse preliminary transformation, the inverse quadratic transformation matrix corresponding to a two-dimensional In the reverse quadratic conversion, one-dimensional inverse quadratic conversion is performed in the specific direction. The video decoding method of claim 21, further comprising: applying another combined reverse conversion perpendicular to the specific direction to the target image data block according to another reverse conversion matrix, wherein the Another inverse conversion matrix is the product of another reverse preliminary conversion matrix and another inverse secondary transformation matrix, and the other reverse The preliminary conversion matrix corresponds to one-dimensional inverse preliminary conversion perpendicular to the specific direction in the two-dimensional inverse preliminary transformation, and the other inverse secondary transformation matrix corresponds to the two-dimensional inverse quadratic transformation perpendicular to the Perform one-dimensional reverse quadratic conversion in a specific direction. The video decoding method of claim 21, wherein the two-dimensional reverse preliminary conversion is the same size as the two-dimensional inverse secondary conversion. The video decoding method according to claim 23 is applied to a digital audio and video codec technical standard (AVS), wherein the specific direction is a horizontal direction, and the inverse conversion matrix is: . The video decoding method according to claim 21, wherein the reverse quadratic conversion matrix is set when the target image data block is determined not to perform one-dimensional inverse secondary conversion in the specific direction. Is a unit matrix. The video decoding method of claim 21, wherein the two-dimensional reverse preliminary conversion is perpendicular to a size of the specific direction greater than a size of the two-dimensional inverse secondary conversion perpendicular to the specific direction; the video decoding method The method further includes: performing a one-dimensional reverse initial conversion in the two-dimensional reverse preliminary conversion on the input image data block before performing the combined back-conversion to generate a first sub-conversion a preliminary conversion result of one of the block and a second sub-block, wherein the first sub-block is the target image data block, and the size of the first sub-block perpendicular to the specific direction is the two-dimensional The inverse quadratic conversion is perpendicular to the size of the specific direction; the second sub-block is subjected to one-dimensional reverse initial conversion in the specific direction in the two-dimensional reverse preliminary conversion; and the target image data area is The block is subjected to the combination of the reverse conversion to produce one of the intermediate turns As a result, a one-dimensional inverse quadratic conversion is performed perpendicular to the specific direction in the two-dimensional inverse quadratic conversion. A video encoding device includes: a conversion module for performing a horizontal preliminary conversion for an image data block: column by column; and immediately starting a horizontal secondary conversion column by column after completing the first column level preliminary conversion; Performing a vertical preliminary conversion by column; and immediately after the initial vertical conversion of the first column, starting a vertical secondary conversion by column; wherein the conversion module starts the vertical preliminary conversion after completing all horizontal secondary conversions. Or start the initial conversion of this level after completing all vertical secondary conversions. For example, the video encoding device described in claim 27 applies a digital audio and video codec technology standard (AVS), wherein the conversion module obtains the following conversion matrix after completing all conversions: A video encoding method is applied to a digital audio and video codec technology standard (AVS), the video encoding method includes: for an image data block: (a) performing a horizontal preliminary conversion column by column; (b) completing the first column Immediately after the initial conversion of the level, a horizontal secondary conversion is started column by column; (c) a vertical preliminary conversion is performed column by column; and (d) immediately after the initial vertical conversion of the first column is completed, a vertical vertical is started column by column. turn Change; wherein step (c) is arranged to start after step (b) is completely completed, or step (a) is arranged to start after step (d) is completely completed. A video decoding device includes: a reverse conversion module for performing a horizontal reverse initial conversion for an image data block; and starting column by column immediately after completing the first column horizontal reverse preliminary conversion a horizontal reverse quadratic conversion; performing a vertical reverse preliminary conversion by column; and, after completing the first vertical vertical reverse initial conversion, immediately starting a vertical reverse secondary conversion by column; wherein the reverse conversion mode The group starts the vertical reverse preliminary conversion after completing all horizontal reverse secondary conversions, or starts the horizontal reverse preliminary conversion after all vertical reverse secondary conversions are completed. For example, the video decoding device described in claim 30 applies a digital audio and video codec technology standard (AVS), wherein the reverse conversion module obtains the following inverse conversion matrix after all conversions are completed: A video decoding method is applied to a digital audio and video codec technology standard (AVS), and the video decoding method comprises: for an image data block: (a) performing a horizontal reverse initial conversion column by column; (b) completing After the initial conversion of the first column of horizontal inversion, a horizontal reverse quadratic conversion is started column by column immediately; (c) performing a vertical reverse preliminary conversion by column; and (d) starting a vertical reverse secondary conversion by column after completing the initial vertical vertical conversion of the first column; wherein step (c) is arranged as Beginning after step (b) is fully completed, or step (a) is scheduled to begin after step (d) is fully completed.